The Future Framework for Diabetes Risk Management Authors: Rupan Bose, Michael J. Wilkinson, Dinesh K. Kalra Published: 2026 ABSTRACT Corresponding Author Cardiovascular disease (CVD) remains the leading cause of death worldwide, and diabetes remains among its most powerful and most rapidly expanding drivers, especially in low- and middle-income countries. An estimated 589 million adults were living with diabetes in 2024, a figure projected to reach 853 million by 2050.1 That diabetes confers roughly a 2-fold excess risk of vascular disease has been established for decades,2 and that excess risk persists in the contemporary treatment era despite substantial gains in lipid and blood pressure control.3 The more consequential questions for prevention is therefore not whether diabetes is a cardiovascular risk factor, but rather when does that risk begin, how does it accumulate, and whether the categorical boundaries we use to define dysglycemia actually capture the biology we are trying to treat. In this issue of JACC: Advances, Domanski et al4 bring a uniquely long lens to that question. Using the Coronary Artery Risk Development in Young Adults (CARDIA) cohort (5,115 Black and White adults enrolled at ages 18-30 years in 1985-1986 and were followed with serial examinations for more than 3 decades), the authors set age 40 years as a landmark, characterized each participant's risk-factor exposure over the preceding 2 decades as the area under the individual trajectory curve, and then followed participants forward for incident CVD. To separate the effect of diabetes from the company it keeps, they constructed a sequential propensity score-matched sample balanced on demographics, lifestyle, body mass index, waist circumference, hypertension history, estimated glomerular filtration rate, and cumulative exposure to low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, and blood pressure, yielding 201 participants with diabetes, 201 with prediabetes, and 364 without either. Over a median of 22 years after age 40 years, 103 first CVD events occurred in the 766 matched participants. Diabetes diagnosed by age 40 years was independently associated with incident CVD (HR: 2.13; 95% CI: 1.33-3.41), coronary heart disease (HR: 2.42; 95% CI: 1.27-4.62), and stroke (HR: 2.37; 95% CI: 1.11-5.06), but not with congestive heart failure. Cumulative glucose exposure from age 18 to 40 years, modeled continuously, was likewise associated with CVD (HR: 1.18 per unit of log-glucose area; 95% CI: 1.08-1.28), with event rates rising monotonically across tertiles of exposure. In an exploratory comparison, the absolute excess CVD rate at age 65 years attributable to diabetes among otherwise well-matched participants (10.3%) exceeded the difference attributable to the constellation of risk factors that typically accompanies diabetes (5.3%). That diabetes is an independent risk factor was confirmatory and not surprising. The authors are to be congratulated for 2 other findings that were important and novel observations. First, in the 565 matched participants who did not meet the criteria for diabetes, greater cumulative glucose exposure before age 40 years remained associated with incident CVD (HR: 1.35; 95% CI: 1.07-1.70). Risk was already accruing below the diagnostic threshold. Second, the categorical label of prediabetes did not discriminate risk once groups were matched (HR: 1.21; 95% CI: 0.71-2.06 vs no diabetes), whereas the continuous measure of glycemic burden did. The exploratory trajectory analysis makes the same point. Participants whose pre-age-40 glucose was both high and rapidly rising had a CVD event rate of 33.3% by age 65 years, compared with 3.3% among those with low and slowly rising glucose. When a category and a continuum are placed side by side, and only the continuum carries signal, the continuum is the better description of the underlying biology. These observations complement prior work linking cumulative glycemic load to cardiovascular risk5 and extend it across a far longer exposure window. This framework will feel familiar to preventive cardiologists, because of the concept of “lower for longer is better” for atherogenic lipoproteins and the cholesterol-years construct, in which an equivalent area accrued earlier in life proved more consequential than the same area accrued later.6 Several other lines of evidence also suggest that glycemic exposure behaves similarly. The 10-year post-trial follow-up of the UK Prospective Diabetes Study showed that a period of early intensive glucose control produced macrovascular benefit that emerged and persisted long after glycemic separation between arms had disappeared, an effect difficult to explain by contemporaneous glycemia alone and consistent with a cumulative “glucose-years” model.7 Meta-analyses also show that younger age at diagnosis of type 2 diabetes is associated with worse vascular outcomes and higher mortality,8 and the TODAY cohort has documented a high rate of complications within 13 years of diagnosis when diabetes starts in youth.9 The recent 2026 American Heart Association/American College of Cardiology/American Diabetes Association/American Society of Nephrology cardiovascular-kidney-metabolic syndrome guideline also emphasizes metabolic and cardiorenal staging across the life course and earlier risk detection.10 The present analysis supplies the missing quantitative anchor: an exposure metric, measured prospectively from young adulthood, that behaves the way that cumulative-exposure paradigms predict for other risk factors like smoking, cholesterol, and blood pressure. Several limitations must, however, be kept in mind. Propensity matching, although the correct tool for the question, retained only 766 of 4,931 participants, and there were only 103 events; CIs are correspondingly wide, and the matched sample is thus less representative than the parent cohort. As in any observational design, residual confounding by unmeasured factors (insulin resistance, adiposity distribution, socioeconomic exposures, treatment intensity) cannot be excluded, and the analysis cannot establish that lowering cumulative glucose exposure lowers risk. The null result for heart failure, based on 34 events, is likely due to low statistical power rather than evidence of absent association, given the well-characterized relationship between diabetes and incident heart failure. The exposure metric itself is expressed as the area under a log-transformed glucose curve in units of log-mg/dL × years, which is not yet interpretable at the bedside. CARDIA participants also lived through an era that preceded sodium-glucose cotransporter-2 inhibitors, glucagon-like peptide-1 receptor agonists, and contemporary lipid therapy, so absolute event rates may not transfer to patients managed today. Finally, the cohort comprised Black and White adults from 4 U.S. communities. Given documented differences in dysglycemia risk across racial and ethnic groups, replication in more diverse populations is necessary before these estimates can be generalized globally. What would it take to make cumulative glycemic burden clinically actionable? Four steps seem essential. First, the metric needs interpretable units and candidate thresholds—a “glucose-years” equivalent expressed in mg/dL × years (or HbA1c-years) above a reference value—so that a clinician can model any given patient on such a curve. Second, its incremental value must be tested against contemporary risk equations, including PREVENT, using reclassification metrics rather than HRs alone; a risk marker that is statistically associated but does not move management is of limited use. Third, the findings must be replicated in cohorts with different ancestries, health system contexts, and treatment eras, ideally through pooled longitudinal studies with repeated glucose measurements. Fourth, and most importantly, the hypothesis must be tested prospectively: does earlier intervention in young adults with rising glycemia below the conventional definition cutoffs alter the cardiovascular trajectory more than interventions that are deferred until the diagnostic threshold is exceeded? Given the decades-long latency demonstrated here, such trials will likely need vascular surrogate endpoints, and they will need to be started in a population we currently do not always treat. Cardiovascular-kidney-metabolic syndrome is an epidemic, and millions of individuals have dysglycemia or prediabetes, so these questions are pressing. Formal studies should now examine the role of metformin and the newer agents (sodium-glucose cotransporter-2 inhibitors or glucagon-like peptide-1 receptor agonists) in the right preventive contexts. Lastly, the fasting glucose threshold of 126 mg/dL is a useful diagnostic convention, historically anchored to where microvascular risk starts to rise. Domanski et al remind us that the vasculature does not heed a threshold calibrated to a different endpoint. If cardiovascular risk begins accruing well before the diagnosis is made, then a framework organized around discrete states—normoglycemia, prediabetes, and diabetes—will systematically underestimate risk in exactly the young patients for whom early intervention has the longest runway. Treating hyperglycemia as a continuum of exposure, defined by both magnitude and duration, aligns diabetes with how we already manage dyslipidemia and hypertension, and it points toward a single unifying prevention goal: flatten the curve early for individuals and, ultimately, for populations. Funding support and author disclosures Dr Bose has received consulting/advisory board fees from Eli Lilly and Company. Dr Wilkinson is a consultant to Amarin, Regeneron, Kaneka, Ionis, and NewAmsterdam; reports advisory board fees from Regeneron, Ionis, Arrowhead, Lilly, and NewAmsterdam; has received grant support from Lilly and Esperion (investigator-initiated studies) and support through an institutional consulting agreement with Novartis paid to his institution for advising on research; and has contracted research grants to his institution with Amgen, Novartis, Ionis, Lilly, Silence, and Mineralys. Dr Kalra has no relationships relevant to the contents of this paper to disclose.